Harrison Wells

Building an AI service biz from the ground up |

[Harrison Wells @wells_harrison]

2.5 months ago, I got honest with myself

I'd been building my biz for 8 months with nothing to show

Empty funnel
No prospects
No word of mouth

So I gave myself 90 days. $20K new revenue or I get a job

Today is day 79

I just signed the biggest deal of my life

Keep pushing

[Harrison Wells @wells_harrison]

Hi, I’m harrison.

I’m 8 months into building a business, $6K in debt, and ready to commit or burn out trying.

I’m committing to $20K in revenue in by Sep 30 + a video a day for all of July.

It’s time to prove I can do this. Who else is building and it’s ride or die?

[Harrison Wells @wells_harrison]

@arshamg_ Move the run into a sandbox and give your team access to create / view runs via a shared webapp.

You can inject your codex subscription auth + mcp env variables into the box, and clone any skills / scripts from git

Happy to chat through it if it would help

[austin petersmith @awwstn]

The Certifiably Insane Way to Build an AI Agent:

1. choose a category where mistake tolerance is roughly the same as it is in self-driving cars. we chose "email-based scheduling assistant." many people want this product, but they immediately fire him if he screws up an interaction with a prospect, a candidate, or a potential investor

2. you learn that the edge cases are too complex and too frequent to be solvable. ours: managing timezones for people who travel (and change travel plans) constantly. knowing when NOT to respond, when to text the customer on the side to verify something, when to follow up, which sub-calendar to use, when to bend the rules on availability, when we can schedule that one type of call during your commute but not the other type of call. sharing your availabilities without compromising your privacy. and on and on.

3. the product doesn't feel viable, but you don't want to give up. you spend hours in a hot tub in Marin with a friend who makes self-driving cars. you make a plan to do it the way they did: hold the steering wheel. you go home and build a human-in-the-loop platform and hire contractors to serve as a backstop and catch mistakes before they happen (and to help design a map of what a world-class EA would do in every weird scenario). you decide trust is the currency in your category, so it must be the thing you won't compromise on. the product must succeed at any scheduling request, no matter how complicated.

4. you instantly feel an overwhelming market pull. so you keep going, growing that team to 75 people working 24/7 to support the nonstop scheduling needs of your customers. tons of engineering time goes to scaling the human platform instead of building the product.

5. you try to raise a Series A and investors say you are insane. your gross margins are extremely negative. they believe this is a problem worth solving, but they don't believe it is as hard to solve as you say. they want AI, not humans. your competitors put "NO HUMANS IN THE LOOP" on their landing pages to call you out. you keep going.

6. you work day and night building the harness that can meet the quality standard your customers have come to expect. you create a massive synthetic gold dataset. audit it, and clean it, label it. repeat. then, experiments. fine-tuning. RL. ACE. DSPy. sub-agents. sub agents for your sub-agents. rebuild the harness. throw more tokens at the problem.

7. some weeks you make big progress. some weeks your evals climb a single basis point, but that's better than nothing. more experiments. more tokens.

john coogan said the hot trend in 2026 will be dogged pursuits. that pushes you to continue the pursuit, doggedly.

8. then, one day, you realize you are scheduling thousands of meetings a day and approaching 50% autopilot with no increase in churn or complaints. you put 150 customers in a full self-driving experiment, and they use the product MORE than they were using it when they had the human backstop. you can really start to let go of the steering wheel.

9. you don't know yet if this was a hill worth climbing, but you are nonetheless stoked that you can see the top. you have created a proprietary map of what to do in a million different situations. nobody else has that map, and the models keep getting better at following maps. your plan was to bet on trust, and your product can be trusted.

today was the first day Howie crossed 50% autopilot:

[Harrison Wells @wells_harrison]

@Teknium @theCTO How do yall handle / recommend working w LinkedIn in Hermes?

[dax @thdxr]

i have seen enough proof now that using a coding agent is a deep skill

it's confusing because the people you see heavily using them produce horrible results

but that's because it's a skill! you can get better and the ceiling seems pretty high - this is very exciting to me

[Harrison Wells @wells_harrison]

@cathrynlavery Woohoo, thank you!! Joining now

[Harrison Wells @wells_harrison]

@sama @rezoundous It’s default is assuming it’s wrong and verifying

Opus assumes it’s right unless you ask it to verify

[Harrison Wells @wells_harrison]

@Shpigford Are you running it on your Mac mini? On a vps?

[Zain Shah @zan2434]

Imagine every pixel on your screen, streamed live directly from a model. No HTML, no layout engine, no code. Just exactly what you want to see.

@eddiejiao_obj, @drewocarr and I built a prototype to see how this could actually work, and set out to make it real. We're calling it Flipbook. (1/5)

[Harrison Wells @wells_harrison]

@ryancarson this is awesome! one of the skills mentioned a "dedicated agent-controlled browser profile"

what do you use for your linkedin browser? agent-browser?

[Harrison Wells @wells_harrison]

@businessbarista i build a lil script that takes a hetzner api key and handles all of this + a few quality of life improvements

https://t.co/G1DNKFDiRz

[Harrison Wells @wells_harrison]

@hwbhatti Lemon

[Harrison Wells @wells_harrison]

@DhravyaShah Awesome, have been waiting for this! Can it handle contextual memory across multiple codebases?

I have 4 active client projects all built across different tech stacks with different preferences, structures, etc

[Harrison Wells @wells_harrison]

@mckaywrigley Is this out yet!? I can’t wait to try it

[Harrison Wells @wells_harrison]

@rabois @sentdefender I don’t feel you’re engaging honestly, but I’ll trade your blurry video with another.

A man was killed in broad daylight. Why is your response to justify it? Will there be a trial? Law and order is the American way, is it not?

[OSINTdefender @sentdefender]

Third angle of today’s shooting of a 37-year-old male by agents with U.S. Immigration and Customs Enforcement (ICE) in Minneapolis, Minnesota, clearly shows one of the agents running away from the scuffle before the shooting carrying the victim's handgun, a Sig P320.

[Harrison Wells @wells_harrison]

I don’t see anything in that video, but for fun, let’s assume he pulled a gun.

When someone is killed, it’s fair (and dare I say, American) to critique, analyze and understand.

Will there be a trial? Will there be any level of accountability?

Someone was just killed in broad daylight, and the immediate response is to justify it and forget. How is that normal?

[Harrison Wells @wells_harrison]

@BoredElonMusk @GinoTheGhost Yes, he is confirmed a citizen

[Harrison Wells @wells_harrison]

@RobertMSterling He didn’t pull a gun though…

[Harrison Wells @wells_harrison]

@rabois @sentdefender Keith, I respect your intelligence highly. You honestly look at this video and it doesn’t worry you in the slightest?

How is that possible?

Truly an honest question looking for an honest response

[Harrison Wells @wells_harrison]

@nateliason @conductor_build Been playing with conductor the past few days, but it seems to ignore my MCP config files. It’s just pulling from the local repo right?